Short answer: both. AI will automate parts of many jobs while creating new work, and the net effect depends on time horizon, industry, and policy. In the near term, expect noticeable job churn (tasks shifting between humans and AI) with displacement concentrated in roles heavy on routine, repeatable information work. Over the longer run, if productivity gains translate into new products, lower prices, and higher demand, AI is likely to be a net job creator—but that outcome isn’t automatic.
What tips the balance
•Task mix: Jobs made of routine, predictable tasks are easier to automate; jobs combining expertise, judgment, and interaction are more likely to be augmented.
•Demand elasticity: If AI makes something much cheaper and quality improves, firms often hire more people to meet higher demand (scale effects).
•Diffusion and complements: New roles arise around data, integration, compliance, security, and change management; these can outgrow the tasks automated away.
•Institutions: Training, safety nets, labor mobility, and competition policy shape whether gains spread widely or concentrate.
Where displacement risk is highest
•Tier-1 customer support, transcription/data entry, rote summarization/report writing, simple bookkeeping, basic QA/testing, routine marketing content, and some entry-level coding tasks.
Where opportunities are growing
•AI-heavy roles: data engineering, ML ops, tooling, evaluation/safety, AI governance and compliance.
•AI-augmented roles: healthcare (documentation, triage support), law (draft review), sales (prospecting, enablement), software (code generation + systems design), education (personalized support), creative work that blends human direction with AI.
•Human-centric work that’s hard to automate: relationship sales, management, trades, care work, field service.
What individuals can do
•Master AI as a tool: build workflows that 2–5x your output/quality in your current domain.
•Double down on human advantages: problem framing, domain expertise, communication, empathy, ethics, and stakeholder management.
•Make your work “AI-resilient”: own outcomes (not just tasks), work across tools/systems, and document measurable productivity/quality gains.
•Keep a learning cadence: small, continuous upskilling beats one-off bootcamps.
What employers can do
•Redesign work around augmentation first; automate only when quality and risk are proven.
•Invest in training, QA, data pipelines, and governance; measure impact on both productivity and error rates.
•Create internal mobility paths so displaced workers can transition into AI-augmented roles.
If you share your industry/role and time horizon, I can give a targeted view of likely risks, new roles to watch, and a 90-day upskilling plan.